RESUMEN
Breast cancer holds the highest diagnosis rate among female tumors and is the leading cause of death among women. Quantitative analysis of radiological images shows the potential to address several medical challenges, including the early detection and classification of breast tumors. In the P.I.N.K study, 66 women were enrolled. Their paired Automated Breast Volume Scanner (ABVS) and Digital Breast Tomosynthesis (DBT) images, annotated with cancerous lesions, populated the first ABVS+DBT dataset. This enabled not only a radiomic analysis for the malignant vs. benign breast cancer classification, but also the comparison of the two modalities. For this purpose, the models were trained using a leave-one-out nested cross-validation strategy combined with a proper threshold selection approach. This approach provides statistically significant results even with medium-sized data sets. Additionally it provides distributional variables of importance, thus identifying the most informative radiomic features. The analysis proved the predictive capacity of radiomic models even using a reduced number of features. Indeed, from tomography we achieved AUC-ROC 89.9 % using 19 features and 92.1 % using 7 of them; while from ABVS we attained an AUC-ROC of 72.3 % using 22 features and 85.8 % using only 3 features. Although the predictive power of DBT outperforms ABVS, when comparing the predictions at the patient level, only 8.7% of lesions are misclassified by both methods, suggesting a partial complementarity. Notably, promising results (AUC-ROC ABVS-DBT 71.8 % - 74.1 % ) were achieved using non-geometric features, thus opening the way to the integration of virtual biopsy in medical routine.
RESUMEN
Breast cancer (BC) has overtaken lung cancer as the most common cancer in the world and the projected incidence rates show a further increase. Early detection through population screening remains the cornerstone of BC control, but a progressive change from early diagnosis only-based to a personalized preventive and risk-reducing approach is widely debated. Risk-stratification models, which also include personal lifestyle risk factors, are under evaluation, although the documentation burden to gather population-based data is relevant and traditional data collection methods show some limitations. This paper provides the preliminary results from the analysis of clinical data provided by radiologists and lifestyle data collected using self-administered questionnaires from 5601 post-menopausal women. The weight of the combinations of women's personal features and lifestyle habits on the BC risk were estimated by combining a model-driven and a data-driven approach to analysis. The weight of each factor on cancer occurrence was assessed using a logistic model. Additionally, communities of women sharing common features were identified and combined in risk profiles using social network analysis techniques. Our results suggest that preventive programs focused on increasing physical activity should be widely promoted, in particular among the oldest women. Additionally, current findings suggest that pregnancy, breast-feeding, salt limitation, and oral contraception use could have different effects on cancer risk, based on the overall woman's risk profile. To overcome the limitations of our data, this work also introduces a mobile health tool, the Dress-PINK, designed to collect real patients' data in an innovative way for improving women's response rate, data accuracy, and completeness as well as the timeliness of data availability. Finally, the tool provides tailored prevention messages to promote critical consciousness, critical thinking, and increased health literacy among the general population.
